We have transistors in the nanometer scale, why couldn’t we create really tiny 8086s?

EDIT: thanks guys I just reinvented the GPU albeit crappier

  • Shadow
    link
    fedilink
    8
    edit-2
    7 months ago

    Because a 8086 is really fucking slow, and writing highly multi threaded code is way more difficult.

    • AmonOP
      link
      fedilink
      47 months ago

      Still in 5yo mode: why couldn’t you run the 8086s at an order of magnitude faster considering they would be much smaller?

      • @squaresinger@lemmy.world
        link
        fedilink
        11 month ago

        Heat generated goes up exponentially compared to the clock speed.

        To explain that more simply: if you run a chip at anything below 1GHz you need no cooling or hardly any cooling at all. Between 1GHz and 2GHz you might get away with passive cooling (a little radiator without a fan). Between 2GHz and 5GHz you need air cooling with a fan blowing air through the radiator. From 5GHz to maybe 6GHz you need a water cooler. From 6GHz to maybe 9GHz you need liquid nitrogen cooling and the world record for a CPU is currently at 9.1GHz. Beyond that is currently not possible, mostly due to the heat generated.

        Also, heat generated equals energy consumed, so pushing CPUs to high performances like that means that they are consuming electricity like crazy. That world record CPU consumes over 1kW alone. That’s about as much as a small space heater.

        Up until the Pentium 4 era, this wasn’t really known, and Intel mainly made their CPUs faster by increasing their clock speed. They actually planned to hit 10GHz a few generations after Pentium 4, but they couldn’t do that and instead they focussed on increasing the CPU performance via other means.

        As a bad analogy: Imagine you want to transport 30 tons of freight. You can use a million bobby cars to transport it. Or you can use one bobby car and run it at 100 times the speed of sound. Or you can use a big semi truck.